[CASSANDRA-19454] Revert switch to approximate time in Dispatcher to avoid mixing with nanoTime() in downstream timeout calculations

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 5.0, 5.1
• Component/s: Messaging/Client
• Bug Category: Availability
• Severity: Low
• Complexity: Low Hanging Fruit
• Discovered By: Code Inspection
• Platform: All
• Since Version: 5.0-alpha1

Source Control Link

https://github.com/apache/cassandra/commit/a26fa6cf2c9a86827d38c448a54878d76bb18e27

Test and Documentation Plan

CI passed (summary attached)

• Assignee: Arun Ganesh
• Reporter: Caleb Rackliffe

Authors

Arun Ganesh

Reviewers

Caleb Rackliffe, David Capwell

• Created: 01/Mar/24 22:47
• Updated: 06/Jun/24 04:34
• Resolved: 07/Mar/24 22:23

Description

CASSANDRA-15241 changed Dispatcher to use the approxTime implementation of MonotonicClock rather than nanoTime(), but clock drift between the two, can potentially cause queries to time out more quickly. We should be able to revert the Dispatcher to use nanoTime() again and similarly change QueriesTable} to {{nanoTime() as well for consistency.

Issue Links

• links to: GH 5.0 PR
• links to: GH trunk PR